NPM 12.1 Sneak Peek - Using Perfstack for Networks

OK, OK, I admit it. My favorite new feature of NPM 12.1 is PerfStack^TM which is actually a feature of the Orion Platform (or “Core”). So my team didn’t actually build it. Darn.  Fortunately, all the products that run on the Orion Platform get it, including NPM.  More importantly, all products can contribute data to it, making it super powerful.

The idea is simple. Put all of the data points that Orion has on a single page and make it pleasantly interactive.  Check it out:

Search for an entity on the left, browse through all the data we have about that entity, and drag and drop it onto the visualization pane on the right.  Works for all sorts of different entities like nodes, interfaces, and volumes.  Works with all sorts of data types too, like status (think: up/down/warning), time-series (think: bandwidth utilization), and events (think: alerts).

What does that look like for networks?  Check it out:

Interface status, transmit utilization and errors, and top egress flows on a single page.  Including NPM and NTA data.  Awesome.  We’ve been wanting to build this into the product for years!  I never would have guessed we would instead build a framework so that YOU can build these views on a whim.

Making it this easy and fast to view data points is absolutely a game changer. I was exploring some of the data the other day.  It’s like the scientific method in real time.  Observe some data, come up with a hypothesis, drag on related data to prove or disprove your hypothesis, rinse and repeat.

Check this out.  We’ve been experiencing slowness on one of our routers here at SolarWinds.  Bandwidth wasn’t saturated, and slowness didn’t seem to be tied to any specific interface.  I started by seeing the performance the poller gets when contacting the router:

That looks good, so why is my transit traffic impacted?  How’s CPU and RAM?  Add those…

RAM is kinda high for a router.  Is that causing any buffer drops?  Yes! Let’s see what size is dropping by pulling in the history for the 5 different sizes.

Ok, so we’ve got high RAM, causing buffer misses, specifically and only for Medium Buffers. I can jump on the router and tell it to be more aggressive about preparing more Medium Buffers to cover spikes of packets of this specific size.

This takes literally seconds, which is exactly what you want during an outage.

And data is provided by everything that works on Orion Core.  Think about that!  Notice high load time on a web page?  Pull in your WPM chart to see the history.  Correlate to CPU usage.  Correlate that to IIS queues.  To interface stats.  To flow data.

PerfStack does several of the things you guys have been asking for, and a few extra things we thought up:

• Imports new data points as they’re polled in near real-time.  Without reloading the page or even a graph.
• Hide the metric panel if you want and PerfStack displays the graph full screen.
• Save your favorite PerfStacks and go back to them whenever you want.
• Send someone the URL of the PerfStack you’re looking at, and they’ll see the same, even without saving.  Of course if they don’t have permission to see some data, they’ll see only the data they have permission to see.
• Update the timeframe and all charts will adhere to it.

Want to try it out for yourself?  Keep an eye on your inbox.  We’ll be emailing you soon that NPM 12.1 is ready in your customer portal!

And what would you do with PerfStack?  We want to hear over at the appropriately titled post: What Would You Do with PerfStack?